Rayid Ghani thinks that data scientists are wasting their talent on meaningless projects. The former chief scientist for the Obama campaign, now based at the University of Chicago, spent years alongside other physicists and mathematicians in pursuit of profits for financial and advertising companies in Silicon Valley and on Wall Street.
"I talked to other people who had the same desire to apply their skills to do something with social impact but didn’t have any idea where to start and how to do it, especially graduate students," says Ghani. After looking for ways to "solve a real problem," he teamed up with Google’s executive chairman Eric Schmidt (whose company also happens to rely on data-heavy search and advertising algorithms) to create the Data Science for Social Good fellowship. Its motto: "We’re training data scientists to tackle the problems that really matter."
The program is working on urban dysfunction as its first target. "Most governments and nonprofits simply don’t know what’s possible yet," explains the program website. "They have data--but often not enough and maybe not the right kind."
The DSSG fellowship has assembled 36 data scientists--"a hybrid of computer scientist, statistician, and domain expert"--in Chicago to trade ad optimization for tackling tough data projects related to education, health, energy and transportation.
DSSG is aiming for some early wins across Chicago’s 28,000 city blocks this summer, and will then expand its scope and geography from there. The first few projects will predict when Chicago’s brand new bikeshare stations will run out of bikes, forecast crime for the Chicago Police Department, warn local hospitals about patients’ impending heart attacks, and gauge the likely performance of energy efficiency loans, among other projects. More are on the way.
These ambitious projects are only part of the solution: the mentality of local, state and federal governments must change for data to have a transformative impact. Traditional policy responses consist of political jockeying and good (or mixed) intentions, fortified with a bit of expert advice and evidence to justify actions. If programs such as Rayid’s succeed, data will move front and center to the core of public policy and designing civic services.
Ghani believes that the Obama campaign showed how the power of datasets is not only to directly influence people and predict events, but also to empower others to take action. Obama’s campaign mobilized an army of volunteers who influenced a large web of social networks that may have tipped the election.
Similarly, governments can use data to give workers and citizens information they need to intervene with individual students in order to slash high school dropout rates--or to take proactive steps to improve their own neighborhoods before problems come up. That type of thinking still remains foreign to most governments.